Protein Design
This challenge focuses on closing the loop between design models and experimental evidence, for target discovery, immune modulation and synthetic circuits.
Read the brief
Structural prediction and generative design tools now produce more candidate binders, enzymes and immune modulators than any lab can test. Most never get past an in-silico confidence score, and the few that do are tested one at a time with assays built for a slower era. This challenge asks teams to build the validation layer: the assays, workflows and decision rules that tell us quickly and cheaply which predictions are real.
Triage at scale
Rank hundreds of designed binders with the cheapest readout that still separates true hits from confident false positives.
From binding to function
Find the early assays that best predict whether a designed immune modulator will work in a cell-based immune assay.
Confidence you can trust
Benchmark model confidence scores against experimental data. When should we believe the model?
Learning from failure
Capture negative results in a standard form so they can feed back into the next round of design.
Designing for the assay
Build validation in from the start. Which design choices make a protein easier to test?
